Re: Dealing with CLUSTER failures

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > I looked over this item, originally posted as:
> > http://archives.postgresql.org/pgsql-hackers/2005-03/msg01055.php
>
> I still think this is substantial complication (more than likely
> introducing some bugs) to deal with a non problem.
>
> http://archives.postgresql.org/pgsql-hackers/2005-03/msg01058.php

Well, it is a demonstrated problem, and we currently don't even report
the index name that caused the cluster to fail.

Here is a new patch that continues to throw an error for a failed
database-wide cluster (no warning) if a single table fails. It does
print the index name and it prints a hint that the user can use ALTER
TABLE ... SET WITHOUT CLUSTER to avoid the failure:

test=> CLUSTER test_gist_idx ON test;
ERROR: cannot cluster on index "test_gist_idx" because access method
does not handle null values
HINT: You may be able to work around this by marking column "a" NOT
NULL.

test=> CLUSTER;
ERROR: cannot cluster on index "test_gist_idx" because access method
does not handle null values
HINT: You may be able to work around this by marking column "a" NOT
NULL, or use ALTER TABLE ... SET WITHOUT CLUSTER to remove the cluster
specification from the table.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073